Re: The original version

In this thread I showed that:
1.
"Anchor Modeling" cannot solve some major areas related to "history" of databases.

2.
A surrogate key is only a special sub case of my solution (see my messages 12 and 34 from this thread). See also at http://www.dbdesign10.com Section 6.3 .
As I mentioned in the above point 1, the surrogate key is a very limited solution.

3.
Some important theoretical solutions are incorrect.

The authors of ''Anchor Modeling'' wrote corrections and published a new version of ''Anchor Modeling'' in December 2010. In the new version of ''Anchor Modeling'' these authors introduced new and very important ideas: states, extensions, binary structures, mappings between data models and identifiers of relationships. I introduced states and identifiers of relationships in my papers from 2005. I introduced decomposition into the binary structures in 2005 and 2006, and extensions and mappings between data models from 2008.

In this post I will concentrate on the following two:

(a) A New Definition of ''historized attribute'':
Definition 7 (Historized Attribute). A historized attribute BH is a string. A historized attribute BH has an anchor A for domain, a data type D for range, and a time type T as time range. An extension of a historized attribute BH is a relation over I x D x T.

This definition degrades the ER and Semantic model. The relation I x D x T is what E.Code unsuccessfully attempted to have in his paper RM/T. He wrote ''In any RM/T databases there is a unary relation (called an E-relation) for each entity type.'' Note that the ''Anchor'' in fact is the ''E-relation''.

Note that the extension from Definition7 enables the following: - the decomposition of data structures into binary structures - spanning a data model over two data models (Conceptual Model and Relational

   Model). It also enables a transition (mapping) from the Conceptual Model to the Relational Model.

Note that the authors of ''Anchor Modeling'' try to solve the mentioned fundamental semantic and structural problems, only by definition7.

In ''Anchor Modeling'' there is no guarantee that (binary structures) ”historized attributes” form an entity or a whole at a given point in time. To be more precise, what is the theoretical basis for creating these binary structures? In RM there are no entities. On May 15, 2006, I presented a solution which explains how to decompose a relation into binary relations. (see Section 4 at http://www.dbdesign10.com ) .

In my paper from 2008 at http://www.dbdesign11.com I presented how to decompose a concept of an entity, a concept of relationship and a concept of a state of an entity or relationship into Binary Concepts. All of this was done on the conceptual level. The binary concept is precisely defined, as well as the conditions under which binary concepts form an entity (see 4.2.2 and 4.2.6 in my paper). In the paper from 2008 I defined mapping from the Conceptual Data Model to the Relational Data Model. This data model mapping is determined by the following two mappings: the schema mapping and data mapping.

The authors of “Anchor Modeling” give no theoretical explanation for surrogate keys. We know that functional dependency is determined by the real state in the Universe of Discourse. However, in this paper there is not a single word about the surrogate keys and the real world.
Note that every ''historized atribute'' has the attribute T. Thus, if an entity has 10
''historized atribute'', then the entity (and the corressponding relation) has 10 of the same attributes T. In my data model, time T is not an attribute.

(b) The Identifiers of Relationships

In this new version from December 2010, the authors of the ''Anchor Modeling'' introduce the identifier of a relationship (see Definition 16).

This definition is a sub case of the identifier of the state of a relationship from my data model (see 4.2.4.1, 4.2.4.2, 4.2.5, 4.2.6, 4.2.7, 4.2.8 and 4.2.9 in my paper from 2008). Note that I presented the main ideas about the identifier of a state of an entity or relationship in my paper from 2005, five years prior to 2010.

Identifiers of states are very important and fundamental.

(i) In my papers I defined a state of an entity or relationship as the
total knowledge about the entity or relationship
(ii) In my paper from 2008 under 3.9 I defined knowledge about an
entity or relationship
(iii) I defined a state as a concept
(iv) The concept of a state of an entity or relationship is a definite
departure from the idea that a concept is determined only by conjunction of properties. This is the first time that one precisely shows the construction of this kind of concepts.
(v) In my papers, states are at the subject level, not at the object
level.
(vi) I distinguish between three realms: the real world i.e. the realm
of reference; the realm of mental; and the realm of semantics. For instance, everyone can grasp a state of an entity in the same way, which means that semantics (meaning and grasping of the mental) is real and objective.
(vii) In my papers knowledge is based on facts. I determined two
important properties of facts: the facts are recorded i.e. they are permanent and the corresponding subjects are aware of the facts. (see 3.4)
(viii) The concepts in my paper work with abstract objects (not with
physical object). In my data model the abstract objects are recorded
(written) in the memory, which is the reason why I put the prefix m in
front of their names. The abstract objects that satisfy the concept of the states are complex because they are constructed from the other abstract objects. This means that depending on their structure, there are two kinds of abstract objects.
(ix) Concerning entity (relationship) changes, the following from my
paper is
significant: there are only two kinds of events related to changes. An event which causes new information and an event which causes existing information to be invalid after this event (this event closes existing information). A very important consequence of my event approach is that any change of a state corresponds to one of these two events in the real world. This approach implies a new definition of and a new approach to time.
(xi) I introduced the identification of a state. Every state has an
identifier. My paper is the first instance in which the identifiers of a state of an entity or relationship were introduced. Note that the states are abstract objects.
(xii) The identifier of a state is determined by the corresponding
concept.
(xiii) The entities involved in relationships must be in the
corresponding states.
(xiv) I introduced a procedure (a), which together with other
constructs resolves
the problem of "history", the semantics of states and changes, and other problems. I described my procedure (a) on this thread in a post from May 26.
(xv) My definition of the concept of a relationship determines the
keys of the relationship.
(xvi) My paper, for the first time, shows how to solve the "history"
of states of an entity or relationship.
(xvii) My paper gives for the first time an effective solution, which
decomposes any data structure of the state into appropriate binary structure. This has been shown for the Conceptual Data Model, Relational Model and File Model.
(xviii) My solution is of a general character; it is at the level of
databases design and concepts. Note that surrogate keys and object identities from OOP have some elements that are on a technical level.

The main ideas for these solutions were presented in 2005 in my papers.

None of the above listed eighteen points is included in the first version of "Anchor Modeling".

In this thread, on 12 June 2010, I criticized the definition of a relationship from the first version of ''anchor modeling" I wrote: ”But in section 2 there is Def2 which says: “Def2. An anchor A(C) is table with one column”. Now it turns out that the relationships between tables with one column
are captured through ties? Of course this is nonsense.” I believe that the paper, which intends to resolve such fundamental issues in database theory, cannot define them this way. In definition13 (relationship) in the new version of ’’Anchor Modeling”, the authors introduce “roles”. Note that Peter Chen defined roles, a long time ago. Note that my definition of a relationship from 2008 also uses also the keys of the relationship (the keys caver roles, see also (xv) in this post).

In this new version from December 2010, the authors of ‘‘Anchor Modeling'' introduce an identifier of a relationship (only by Definition 16).

Note that the identifier of state is a vital part of the following:

1. the solution for ”history”
2. the decomposition the data structures into binary structures
3. It completely defines the mapping between data models
4. It identifies the states
5. It identifies abstract complex objects,

Vladimir Odrljin Received on Wed Jul 06 2011 - 22:58:53 CEST